Improving Political Discourse Analysis on Twitter With Context Analysis

Aritz Bilbao-Jayo,Aitor Almeida

doi:10.1109/access.2021.3099093

Abstract

In this study, we propose a new approach to perform political discourse analysis in social media platforms based on a widely used political categorisation schema in the field of political science, namely, the Comparative Manifestos Project's category schema. This categorisation schema has been traditionally used to perform content analysis in political manifestos, giving a code that indicates the domain or category of each of the phrases in the manifestos. Therefore, in this work we propose the application of this political discourse analysis technique in Twitter, using as training data of 100 publicly available annotated political manifestos in English with around 85,000 annotated sentences. Furthermore, we also analyse the improvement that using 5,000 annotated tweets could provide to the performance of the political discourse classifier already trained with political manifestos. Finally, we have analysed the 2016 United States presidential elections on Twitter using the proposed approach. As our main finding, we have been able to conclude that both datasets (political manifestos and annotated tweets) can be combined in order to achieve better results, achieving improvements in the F-Measure of more than 15 points. Moreover, we have also analysed if contextual information such as the previous tweet or the political affiliation of the transmitter could improve classifier's performance as it has already been proven for manifestos classification, introducing a novel method for political parties representation and finding that adding the previous tweet or the political leaning as contextual data does improve its performance.

Highlights

T HE rise of social media have offered both politicians and citizens new ways of interacting directly with each other without the direct mediation of traditional media
In this study, we propose a new approach to perform political discourse analysis in social media platforms based on a widely used political categorisation schema in the field of political science, namely, the Comparative Manifestos Project’s category schema
Regarding how the previous phrase has been added to the model as a new input in order to improve the performance of the model as it will be demonstrated in Section IV, we have replicated for the previous tweet, the same convolutionpooling process it is used in the actual tweet that is going to be analysed

Summary

Introduction

T HE rise of social media have offered both politicians and citizens new ways of interacting directly with each other without the direct mediation of traditional media This phenomenon has allowed citizens to become participants in the construction of the political agenda, forcing political parties to use more direct means of communication than the mainstream press and media [1]. As all the members of these on-line social media platforms are treated as equals, any citizen can send a message to the politician, sometimes leading to a discussion between the politician and the citizens or between citizens themselves These social media platforms contain valuable data regarding citizens’ concerns or the politicians’ current talking points. As Dimitrova and Matthes [2] stated in their work analysing how social media behaves in Political Campaigning, social media have become an indispensable part of modern political campaigning, both in the United States and internationally

Objectives

Results

Conclusion